Ethnicity & Health
ISSN: 1355-7858 (Print) 1465-3419 (Online) Journal homepage: http://www.tandfonline.com/loi/ceth20
The influence of ethnicity on the outcomes of violence in pregnancy Pauline Gulliver & Robyn Dixon To cite this article: Pauline Gulliver & Robyn Dixon (2015) The influence of ethnicity on the outcomes of violence in pregnancy, Ethnicity & Health, 20:5, 511-522, DOI: 10.1080/13557858.2014.939577 To link to this article: http://dx.doi.org/10.1080/13557858.2014.939577
Published online: 23 Jul 2014.
Submit your article to this journal
Article views: 206
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ceth20 Download by: [Australian National University]
Date: 05 November 2015, At: 15:50
Ethnicity & Health, 2015 Vol. 20, No. 5, 511–522, http://dx.doi.org/10.1080/13557858.2014.939577
The influence of ethnicity on the outcomes of violence in pregnancy Pauline Gullivera* and Robyn Dixonb School of Population Health, University of Auckland, Auckland, New Zealand; bSchool of Nursing, University of Auckland, Auckland, New Zealand
Downloaded by [Australian National University] at 15:50 05 November 2015
a
(Received 4 February 2014; accepted 25 June 2014) Aim. To investigate the influence of ethnicity on immediate and long-term (five years post-partum) foetal, maternal and injury-related outcomes. Methods. A retrospective, population-based study analysing maternal discharge records linked to birth/death certificates for women aged 25 years and under. Discharge records were grouped according to ethnicity (Maori and non-Maori) and as follows: (1) a pregnancy-related hospital admission, but no associated or subsequent assault recorded (pregnant only); (2) an assault-related hospital admission event after the pregnancy, but within five years of the index pregnancy (assault after pregnancy); and (3) an assault recorded within the same hospital admission event as the pregnancy (assault during pregnancy). Generalised linear models for the binomial family were conducted to explore increased risk ratios of pregnancy-related and subsequent injury outcomes depending on ethnicity and group assignment. Results. Compared with the pregnancy-only group, rate ratios (RRs) for maternal and foetal outcomes were higher in the assault after pregnancy group and the assault during pregnancy group. For injury outcomes in the five years after the injury event, RRs for the assault after pregnancy group exceeded both the pregnancy-only and the assault during pregnancy groups. RRs for non-Maori women assaulted after pregnancy were higher for injury hospitalisations, fracture and intracranial injury than those for Maori women. Conclusion. Given that Maori women experience a higher prevalence of severe intimate partner violence and more difficulties accessing health care, we suggest that the findings highlight potential problems for health care access for Maori women experiencing violence. Keywords: violence against women; ethnicity; health care access
Background The 2012 New Zealand Human Rights Commission’s report to the Committee on the Elimination of Discrimination Against Women highlighted the continual problems that New Zealand wrestles with in reducing violence against women (McGregor 2012). Indigenous (Maori) women are at particularly high risk of sexual and family violence. Although Maori comprise approximately 15% of the New Zealand population, just under one-third of sexual violence victims are Maori (McGregor 2012). In addition, the rates of family violence death for Maori victims are four times that of ‘other’ ethnicities (Family Violence Death Review Committee 2013), while 42% of women who accessed
*Corresponding author. Email: [email protected] © 2014 Taylor & Francis
Downloaded by [Australian National University] at 15:50 05 November 2015
512
P. Gulliver and R. Dixon
the services of the New Zealand National Collective of Women’s Refuges in 2013 were Maori (National Collective of Independent Women’s Refuges Inc. 2013). Marie, Fergusson, and Boden (2008) caution against suggesting a causal link between ethnicity and violence, suggesting that it is a unique interplay of sociocultural, personal and situational factors that increase vulnerability (Marie, Fergusson, and Boden 2008). As with indigenous people from other countries (Trocme, Knoke, and Blackstock 2004), Maori are over-represented in low socio-economic groups (New Zealand Child and Youth Epidemiology Service 2012) and in unemployment statistics (Statistics New Zealand 2013c). Attempts to disentangle the effects of socio-economic status and indigenous status in New Zealand and internationally show that although adjustment for socioeconomic and familial cofounders attenuate relationships between indigenous status and violence experience, the odds of experiencing violence continue to be elevated (Marie, Fergusson, and Boden 2008; Daoud et al. 2013). It has been argued that colonisation theory underscores the historical roots of increased vulnerability for indigenous populations. To this end, Daoud and colleagues identify three elements of colonisation theory that help to explain the high levels of violence experienced by indigenous women: (1) Collective violence: cultural discrimination and violations of human rights (Razack 1994); (2) Shifts in balance and power: the imposition of colonial values that destroyed balanced power relationships between men and women (Stevenson 2011); (3) The impact of colonial policies: including the forced removal of indigenous children from their families and the degradation of indigenous language (Ing 2011). While previous research has facilitated our understanding that differences in indigenous and non-indigenous experiences of intimate partner violence (IPV) cannot be solely explained by socio-economic status, and that historical factors must be acknowledged, indigenous women are also exposed to other risk factors for IPV that have not been fully explored. For example, compared to the total New Zealand population, the Maori population of New Zealand is younger [median age, Maori population: 23.9 years; median age, total population: 38.0 years (Statistics New Zealand 2013a, 2013b)], and more than one quarter of dependent children in New Zealand live in Maori households (Te Puni Kokiri 2011). Younger age is an established risk factor for exposure to IPV (Chu, Goodwin, and D’Angelo 2010), as is multiparity (Taillieu and Brownridge 2010). A considerable number of women experience IPV for the first time during pregnancy. Previous research indicates that pregnancy may be a time when violence escalates, or one when the nature of violence changes from physical to emotional and/or sexual abuse (Taillieu and Brownridge 2010) with Canadian (Muhajarine and D’Arcy 1999) and New Zealand (Fanslow et al. 2008) research reporting increased risks for indigenous populations compared to non-indigenous populations. Further, violence in pregnancy has also been associated with adverse outcomes for both the mother and the child (El Kady et al. 2005), as well as increased risks of subsequent physical abuse in the 12 months post-partum (Gartland et al. 2011). In this investigation, we were interested in exploring the influence of ethnicity on immediate and long-term (five years post-partum) foetal, maternal and injury-related outcomes. By limiting our investigation to women aged 25 years or under, we attempted
Ethnicity & Health
513
to account for the different age structure between Maori and non-Maori mothers. Overall, the Maori population is younger than the non-Maori population of New Zealand (Statistics New Zealand 2013a, 2013b), and Maori women begin having children at a younger age [fertility rates are highest in Maori women between age 20 and 29 at 150 births per 1000 women, while for non-Maori women, the fertility rates are highest at age 30–34 years at 125 births per 1000 women (Statistics New Zealand 2011)].
Downloaded by [Australian National University] at 15:50 05 November 2015
Methods The methods used in this investigation are based on a retrospective, population-based study analysing maternal discharge records linked to birth/death certificates from 1991 to 1999, described by El Kady et al. (2005). Approval to undertake the investigation was granted by the University of Auckland Human Participants Ethics Committee (Ref: 8404). Data used in this investigation were collected and maintained by the Ministry of Health (MoH) who use the International Classification of Diseases and Other Health Related Problems, version 10, Australian Modification (ICD-10-AM) to record the nature and circumstances surrounding a hospital event (National Centre for Classification in Health 2002b). All hospitalisation assaults during pregnancy that were recorded are captured in the MoH National Minimum Data Set of hospital discharges (NMDS). It has been estimated that, in New Zealand, over 99% of injury-related hospital events are publicly funded, thus the NMDS offers an accurate representation of hospitalised injuries (Davie et al. 2008). It should be noted that home births are not included in the NMDS. In 2010, these accounted for 3.2% of births (Ministry of Health 2012b). For the purposes of the study, hospital events for women aged 25 years and under, who were admitted to a New Zealand public hospital with an ICD-10-AM diagnosis code in the range O00–O99 (pregnancy, childbirth and the puerperium) or Z37 (outcome of delivery) for the period 2001–2006, were identified and included in the analysis. Pregnancies that result in delivery before 20 weeks of gestation are not recorded in the NMDS. National Health Index numbers (unique identifiers assigned to specific individuals and used consistently across MoH data-sets) were used to identify the index pregnancy event and subsequent discharge events for the five years after the index event. Only the first pregnancy-related admission for each woman that occurred between 2001 and 2006 was considered to ensure that five years of follow-up hospital discharge information was available for all pregnancies included in this period. The women were assigned to one of three groups depending upon the information contained within their hospital discharge records: (1) Pregnant only: Those women for whom there was a pregnancy-related hospital admission, but no associated or subsequent assault was recorded. (2) Assault after pregnancy: Those who had an assault-related hospital admission event after the pregnancy, but within five years of the index pregnancy. (3) Assault during pregnancy: Those who had an assault recorded within the same hospital admission event as the pregnancy. Assault-related hospital events were those with an ICD-10-AM external cause code in the range X85–Y09.
Downloaded by [Australian National University] at 15:50 05 November 2015
514
P. Gulliver and R. Dixon
We intended to investigate the same maternal and foetal outcome measures as those reported by El Kady et al. (2005); however, small numbers precluded the investigation of a number of outcomes.1 The MoH mapping algorithms were used to convert El Kady et al.’s ICD-9 codes to ICD-10-AM codes. The maternal outcomes measured and their associated ICD-10-AM codes were: preterm labour (O47.0, O60); premature rupture of the membranes (O42); and antepartum haemorrhage (O46). The only foetal outcome that could be included in this investigation was stillbirth (Z37.1, Z37.3, Z37.4, Z37.6 and Z37.7). Subsequent injury outcomes (in the five years following the pregnancy admission) were also recorded to allow comparison with El Kady et al. (2005), including: total injuries (S00-T78), fractures (S00-T13 diagnosis codes, third character = 2, plus T142), intracranial injuries (ICD-10-AM three-character diagnosis code = S06) and open wounds (S00-T02 diagnosis codes, third character = 1, plus T091, T11.1, T13.1 and T14.1).
Ethnicity Any one hospital admission record may include up to three self-reported ethnic groups (Ministry of Health 2009). We used ‘prioritised reporting’ for identifying ethnicity from the hospital records. In prioritised reporting, if Maori is one of the groups reported, then ethnicity is assigned to Maori. Otherwise, if any Pacific Island group is reported, then ethnicity is Pacific Island; then if Asian is reported then ethnicity is Asian; then if any group other than European is reported, then ethnicity is assigned to that group; then assign to other. Using this reporting scheme, priority is given to non-European groups, but special priority is given to New Zealand Maori and Pacific Island groups (Statistics New Zealand 1997).
Analysis Data analysis was conducted using Stata11SE (1985; StataCorp). Chi-squared analyses were conducted to determine differences in demographic variables and length of hospital stay between Maori and non-Maori mothers. Generalised linear models for the binomial family were conducted to explore risk ratios of each of the outcomes depending on group assignment. Each outcome was tested using an ever/never distinction. Models were analysed separately for Maori and non-Maori mothers to identify differences in outcomes by ethnicity. We also investigated the effect of adjusting the models for the number of live deliveries in the five-year follow-up period and deprivation. Risk ratios and 95% confidence intervals (CI) are reported.
Results The demographic and hospital admission characteristics of the cohort are presented in Table 1. A lower proportion of Maori mothers were living in areas of low economic deprivation than non-Maori mothers. Also a lower proportion of Maori mothers spent four or more days in hospital following the index pregnancy event, had no additional live deliveries in the five years after the event and were in the ‘pregnancy-only’ comparison group. The following ethnicities were reported by those in the non-Maori group:
Ethnicity & Health
515
Table 1. Demographic and hospital admission characteristics of cohort. Total cohort
Downloaded by [Australian National University] at 15:50 05 November 2015
N
%
Non-Maori N
%
Maori N
%
Age group 6 days Total
20,611 37.5 27,688 50.4 5340 9.7 1341 2.4 54,980
13,543 37.8 17,704 49.4 3713 10.4 855 2.4 35,815
7068 36.9 9984 52.1 1627 8.5 486 2.5 19,165
Number of live deliveries in five years after index event 0 11,443 20.8 9172 25.6 1 24,295 44.2 16,228 45.3 >1 19,242 35.0 10,415 29.1 Total 54,980 35,815
2271 11.8 8067 42.1 8827 46.1 19,165
Comparison group Pregnancy only Assault after pregnancy Assault during pregnancy
53,805 97.9 958 1.7 217 0.4
35,297 98.6 431 1.2 87 0.2
18,508 96.6 527 2.7 130 0.7
Chi-squared
p
1.29
0.26
1800
ISSN: 1355-7858 (Print) 1465-3419 (Online) Journal homepage: http://www.tandfonline.com/loi/ceth20
The influence of ethnicity on the outcomes of violence in pregnancy Pauline Gulliver & Robyn Dixon To cite this article: Pauline Gulliver & Robyn Dixon (2015) The influence of ethnicity on the outcomes of violence in pregnancy, Ethnicity & Health, 20:5, 511-522, DOI: 10.1080/13557858.2014.939577 To link to this article: http://dx.doi.org/10.1080/13557858.2014.939577
Published online: 23 Jul 2014.
Submit your article to this journal
Article views: 206
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ceth20 Download by: [Australian National University]
Date: 05 November 2015, At: 15:50
Ethnicity & Health, 2015 Vol. 20, No. 5, 511–522, http://dx.doi.org/10.1080/13557858.2014.939577
The influence of ethnicity on the outcomes of violence in pregnancy Pauline Gullivera* and Robyn Dixonb School of Population Health, University of Auckland, Auckland, New Zealand; bSchool of Nursing, University of Auckland, Auckland, New Zealand
Downloaded by [Australian National University] at 15:50 05 November 2015
a
(Received 4 February 2014; accepted 25 June 2014) Aim. To investigate the influence of ethnicity on immediate and long-term (five years post-partum) foetal, maternal and injury-related outcomes. Methods. A retrospective, population-based study analysing maternal discharge records linked to birth/death certificates for women aged 25 years and under. Discharge records were grouped according to ethnicity (Maori and non-Maori) and as follows: (1) a pregnancy-related hospital admission, but no associated or subsequent assault recorded (pregnant only); (2) an assault-related hospital admission event after the pregnancy, but within five years of the index pregnancy (assault after pregnancy); and (3) an assault recorded within the same hospital admission event as the pregnancy (assault during pregnancy). Generalised linear models for the binomial family were conducted to explore increased risk ratios of pregnancy-related and subsequent injury outcomes depending on ethnicity and group assignment. Results. Compared with the pregnancy-only group, rate ratios (RRs) for maternal and foetal outcomes were higher in the assault after pregnancy group and the assault during pregnancy group. For injury outcomes in the five years after the injury event, RRs for the assault after pregnancy group exceeded both the pregnancy-only and the assault during pregnancy groups. RRs for non-Maori women assaulted after pregnancy were higher for injury hospitalisations, fracture and intracranial injury than those for Maori women. Conclusion. Given that Maori women experience a higher prevalence of severe intimate partner violence and more difficulties accessing health care, we suggest that the findings highlight potential problems for health care access for Maori women experiencing violence. Keywords: violence against women; ethnicity; health care access
Background The 2012 New Zealand Human Rights Commission’s report to the Committee on the Elimination of Discrimination Against Women highlighted the continual problems that New Zealand wrestles with in reducing violence against women (McGregor 2012). Indigenous (Maori) women are at particularly high risk of sexual and family violence. Although Maori comprise approximately 15% of the New Zealand population, just under one-third of sexual violence victims are Maori (McGregor 2012). In addition, the rates of family violence death for Maori victims are four times that of ‘other’ ethnicities (Family Violence Death Review Committee 2013), while 42% of women who accessed
*Corresponding author. Email: [email protected] © 2014 Taylor & Francis
Downloaded by [Australian National University] at 15:50 05 November 2015
512
P. Gulliver and R. Dixon
the services of the New Zealand National Collective of Women’s Refuges in 2013 were Maori (National Collective of Independent Women’s Refuges Inc. 2013). Marie, Fergusson, and Boden (2008) caution against suggesting a causal link between ethnicity and violence, suggesting that it is a unique interplay of sociocultural, personal and situational factors that increase vulnerability (Marie, Fergusson, and Boden 2008). As with indigenous people from other countries (Trocme, Knoke, and Blackstock 2004), Maori are over-represented in low socio-economic groups (New Zealand Child and Youth Epidemiology Service 2012) and in unemployment statistics (Statistics New Zealand 2013c). Attempts to disentangle the effects of socio-economic status and indigenous status in New Zealand and internationally show that although adjustment for socioeconomic and familial cofounders attenuate relationships between indigenous status and violence experience, the odds of experiencing violence continue to be elevated (Marie, Fergusson, and Boden 2008; Daoud et al. 2013). It has been argued that colonisation theory underscores the historical roots of increased vulnerability for indigenous populations. To this end, Daoud and colleagues identify three elements of colonisation theory that help to explain the high levels of violence experienced by indigenous women: (1) Collective violence: cultural discrimination and violations of human rights (Razack 1994); (2) Shifts in balance and power: the imposition of colonial values that destroyed balanced power relationships between men and women (Stevenson 2011); (3) The impact of colonial policies: including the forced removal of indigenous children from their families and the degradation of indigenous language (Ing 2011). While previous research has facilitated our understanding that differences in indigenous and non-indigenous experiences of intimate partner violence (IPV) cannot be solely explained by socio-economic status, and that historical factors must be acknowledged, indigenous women are also exposed to other risk factors for IPV that have not been fully explored. For example, compared to the total New Zealand population, the Maori population of New Zealand is younger [median age, Maori population: 23.9 years; median age, total population: 38.0 years (Statistics New Zealand 2013a, 2013b)], and more than one quarter of dependent children in New Zealand live in Maori households (Te Puni Kokiri 2011). Younger age is an established risk factor for exposure to IPV (Chu, Goodwin, and D’Angelo 2010), as is multiparity (Taillieu and Brownridge 2010). A considerable number of women experience IPV for the first time during pregnancy. Previous research indicates that pregnancy may be a time when violence escalates, or one when the nature of violence changes from physical to emotional and/or sexual abuse (Taillieu and Brownridge 2010) with Canadian (Muhajarine and D’Arcy 1999) and New Zealand (Fanslow et al. 2008) research reporting increased risks for indigenous populations compared to non-indigenous populations. Further, violence in pregnancy has also been associated with adverse outcomes for both the mother and the child (El Kady et al. 2005), as well as increased risks of subsequent physical abuse in the 12 months post-partum (Gartland et al. 2011). In this investigation, we were interested in exploring the influence of ethnicity on immediate and long-term (five years post-partum) foetal, maternal and injury-related outcomes. By limiting our investigation to women aged 25 years or under, we attempted
Ethnicity & Health
513
to account for the different age structure between Maori and non-Maori mothers. Overall, the Maori population is younger than the non-Maori population of New Zealand (Statistics New Zealand 2013a, 2013b), and Maori women begin having children at a younger age [fertility rates are highest in Maori women between age 20 and 29 at 150 births per 1000 women, while for non-Maori women, the fertility rates are highest at age 30–34 years at 125 births per 1000 women (Statistics New Zealand 2011)].
Downloaded by [Australian National University] at 15:50 05 November 2015
Methods The methods used in this investigation are based on a retrospective, population-based study analysing maternal discharge records linked to birth/death certificates from 1991 to 1999, described by El Kady et al. (2005). Approval to undertake the investigation was granted by the University of Auckland Human Participants Ethics Committee (Ref: 8404). Data used in this investigation were collected and maintained by the Ministry of Health (MoH) who use the International Classification of Diseases and Other Health Related Problems, version 10, Australian Modification (ICD-10-AM) to record the nature and circumstances surrounding a hospital event (National Centre for Classification in Health 2002b). All hospitalisation assaults during pregnancy that were recorded are captured in the MoH National Minimum Data Set of hospital discharges (NMDS). It has been estimated that, in New Zealand, over 99% of injury-related hospital events are publicly funded, thus the NMDS offers an accurate representation of hospitalised injuries (Davie et al. 2008). It should be noted that home births are not included in the NMDS. In 2010, these accounted for 3.2% of births (Ministry of Health 2012b). For the purposes of the study, hospital events for women aged 25 years and under, who were admitted to a New Zealand public hospital with an ICD-10-AM diagnosis code in the range O00–O99 (pregnancy, childbirth and the puerperium) or Z37 (outcome of delivery) for the period 2001–2006, were identified and included in the analysis. Pregnancies that result in delivery before 20 weeks of gestation are not recorded in the NMDS. National Health Index numbers (unique identifiers assigned to specific individuals and used consistently across MoH data-sets) were used to identify the index pregnancy event and subsequent discharge events for the five years after the index event. Only the first pregnancy-related admission for each woman that occurred between 2001 and 2006 was considered to ensure that five years of follow-up hospital discharge information was available for all pregnancies included in this period. The women were assigned to one of three groups depending upon the information contained within their hospital discharge records: (1) Pregnant only: Those women for whom there was a pregnancy-related hospital admission, but no associated or subsequent assault was recorded. (2) Assault after pregnancy: Those who had an assault-related hospital admission event after the pregnancy, but within five years of the index pregnancy. (3) Assault during pregnancy: Those who had an assault recorded within the same hospital admission event as the pregnancy. Assault-related hospital events were those with an ICD-10-AM external cause code in the range X85–Y09.
Downloaded by [Australian National University] at 15:50 05 November 2015
514
P. Gulliver and R. Dixon
We intended to investigate the same maternal and foetal outcome measures as those reported by El Kady et al. (2005); however, small numbers precluded the investigation of a number of outcomes.1 The MoH mapping algorithms were used to convert El Kady et al.’s ICD-9 codes to ICD-10-AM codes. The maternal outcomes measured and their associated ICD-10-AM codes were: preterm labour (O47.0, O60); premature rupture of the membranes (O42); and antepartum haemorrhage (O46). The only foetal outcome that could be included in this investigation was stillbirth (Z37.1, Z37.3, Z37.4, Z37.6 and Z37.7). Subsequent injury outcomes (in the five years following the pregnancy admission) were also recorded to allow comparison with El Kady et al. (2005), including: total injuries (S00-T78), fractures (S00-T13 diagnosis codes, third character = 2, plus T142), intracranial injuries (ICD-10-AM three-character diagnosis code = S06) and open wounds (S00-T02 diagnosis codes, third character = 1, plus T091, T11.1, T13.1 and T14.1).
Ethnicity Any one hospital admission record may include up to three self-reported ethnic groups (Ministry of Health 2009). We used ‘prioritised reporting’ for identifying ethnicity from the hospital records. In prioritised reporting, if Maori is one of the groups reported, then ethnicity is assigned to Maori. Otherwise, if any Pacific Island group is reported, then ethnicity is Pacific Island; then if Asian is reported then ethnicity is Asian; then if any group other than European is reported, then ethnicity is assigned to that group; then assign to other. Using this reporting scheme, priority is given to non-European groups, but special priority is given to New Zealand Maori and Pacific Island groups (Statistics New Zealand 1997).
Analysis Data analysis was conducted using Stata11SE (1985; StataCorp). Chi-squared analyses were conducted to determine differences in demographic variables and length of hospital stay between Maori and non-Maori mothers. Generalised linear models for the binomial family were conducted to explore risk ratios of each of the outcomes depending on group assignment. Each outcome was tested using an ever/never distinction. Models were analysed separately for Maori and non-Maori mothers to identify differences in outcomes by ethnicity. We also investigated the effect of adjusting the models for the number of live deliveries in the five-year follow-up period and deprivation. Risk ratios and 95% confidence intervals (CI) are reported.
Results The demographic and hospital admission characteristics of the cohort are presented in Table 1. A lower proportion of Maori mothers were living in areas of low economic deprivation than non-Maori mothers. Also a lower proportion of Maori mothers spent four or more days in hospital following the index pregnancy event, had no additional live deliveries in the five years after the event and were in the ‘pregnancy-only’ comparison group. The following ethnicities were reported by those in the non-Maori group:
Ethnicity & Health
515
Table 1. Demographic and hospital admission characteristics of cohort. Total cohort
Downloaded by [Australian National University] at 15:50 05 November 2015
N
%
Non-Maori N
%
Maori N
%
Age group 6 days Total
20,611 37.5 27,688 50.4 5340 9.7 1341 2.4 54,980
13,543 37.8 17,704 49.4 3713 10.4 855 2.4 35,815
7068 36.9 9984 52.1 1627 8.5 486 2.5 19,165
Number of live deliveries in five years after index event 0 11,443 20.8 9172 25.6 1 24,295 44.2 16,228 45.3 >1 19,242 35.0 10,415 29.1 Total 54,980 35,815
2271 11.8 8067 42.1 8827 46.1 19,165
Comparison group Pregnancy only Assault after pregnancy Assault during pregnancy
53,805 97.9 958 1.7 217 0.4
35,297 98.6 431 1.2 87 0.2
18,508 96.6 527 2.7 130 0.7
Chi-squared
p
1.29
0.26
1800
